VMD模糊熵和SVM在柱塞泵故障诊断中的应用

Application of VMD Fuzzy Entropy and SVM in Plunger Pump Fault Diagnosis

为有效提取非线性非平稳特性的柱塞泵故障特征,提高故障诊断准确率,提出了一种基于变分模态分解(Variational Mode Decomposition,VMD)模糊熵和支持向量机(Vupport Vector Machine,SVM)相结合的柱塞泵故障诊断方法。首先将信号经过VMD分解形成K个固有模态分量(Intrinsic Modal Component,IMF);然后确定IMF个数,提出了基于峭度分析的IMF个数确定方法;其次取峭度值较大的IMF并计算其模糊熵,确定了各状态下相应的模糊熵;最后将模糊熵作为特征向量输入SVM进行故障识别,准确率可达98.3%。将该方法与经验模态分解(Empirical Mode Decomposition,EMD)模糊熵-SVM、VMD模糊熵-BP神经网络对比,结果表明,VMD模糊熵和SVM相结合的方法在柱塞泵故障诊断中具有优越性。

In order to effectively extract the fault characteristics of the plunger pump with nonlinear and non-stationary characteristics and improve the accuracy of fault diagnosis,a variational mode decomposition(VMD)fuzzy entropy and support vector machine(SVM)Combined plunger pump fault diagnosis method.First,the signal is decomposed by VMD to form K Intrinsic modal components(IMF);then the number of IMFs is determined,and a method for determining the number of IMFs based on kurtosis analysis is proposed;secondly,the IMF with a larger kurtosis value is taken and Calculate the fuzzy entropy and determine the coresponding fuzzy entropy in each state.Finally,the fuzzy entropy is used as the feature vector to input SVM for fault identification,and the accuracy rate can reach 98.3%.Comparing this method with EMD fuzzy entropy-SVM and VMD fuzzy entropy-BP neural networks,the results show that the method combining VMD fuzzy entropy and SVM is superior in plunger pumpfaultdiagnosis.